Using data science to facilitate the delivery of value-based care.

Abstract

Introduction: Unexpected readmission to hospital is a burden to patients, hospitals and funders. At Epworth a wealth of data is accessible through several data sources; yet these data are typically underutilised raw assets. Data science techniques enable leveraging such data to enhance quality and reduce cost of care. This project investigates the potential of data science to predict patients at risk of readmission. Methodology: This research is exploratory in nature using a multiple case study approach. Case study is the appropriate research method in this instance as it allows the researcher to obtain the insider view of the selected case scenario, thus enabling a better understanding of the current clinical status directly from informants within the particular case (Fletcher and Plakoyiannaki 2011; Yin 2003). The chosen cases focus particularly on urology and orthopaedic. The study consists of 6 key phases; 1) Understanding the structures of the various data sets, 2) Data wrangling, 3) Modelling, 4) Evaluation, 5) Deployment and 6) Final evaluation and feedback exchange. Discussion and conclusions: This research has multiple benefits: First, the developed model will assist with reducing unexpected readmissions that can be implemented throughout Epworth. Such a solution, once implemented will enable Epworth to enhance patient wellbeing, avoid unnecessary costs, and support negotiating position with funding bodies. Furthermore, being able to predict risk factors for avoidable/unplanned readmissions will ensure that at all times it will be possible to deliver high quality